A system and associated methodology recognizes an Arabic like alphanumeric character using fuzzy modeling. The method receives a handwritten Arabic like alphanumeric character, stores fuzzy models of a plurality of Arabic like alphanumeric characters, preprocesses the handwritten Arabic like alphanumeric character, extracts features from the preprocessed Arabic like alphanumeric character, computes a similarity value based on fuzzy comparison between points of the preprocessed Arabic like alphanumeric character and the stored fuzzy models, classifies the handwritten Arabic like alphanumeric character based at least in part on the similarity value and outputs a classified alphanumeric character.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method for Arabic like alphanumeric character recognition, comprising: receiving a handwritten Arabic like alphanumeric character, from an alphanumeric input device; storing, in a memory, fuzzy models of a plurality of Arabic like alphanumeric characters; preprocessing, by processing circuitry, the handwritten Arabic like alphanumeric character; extracting, by the processing circuitry, features from the preprocessed Arabic like alphanumeric character; computing, by the processing circuitry, a similarity value based on fuzzy comparisons between points of the preprocessed Arabic like alphanumeric character and the stored fuzzy models; classifying, by the processing circuitry, the handwritten Arabic like alphanumeric character based at least in part on the similarity value; and outputting a classified alphanumeric character, wherein the similarity value between a preprocessed Arabic like alphanumeric character S and a fuzzy model M is calculated based on a membership value calculated by applying sim ( S , M ) = 1 N ∑ i = 1 N M v ( S ( i ) , M ( i ) ) wherein the membership value is expressed as: M v ( S ( i ) , M ( i ) ) = { 0 , d > α 1 or d < α 2 1 , β 1 ≤ d ≤ β 2 α 1 - d α 1 - β , β 1 < d ≤ α 1 d - α 2 β 2 - α 2 , α 2 ≤ d < β 2 where N is a number of points, d is a directional feature of the preprocessed Arabic like alphanumeric character S at point i, β 1 and β 2 are a first tolerance and α 1 and α 2 are a second tolerance.
2. The method of claim 1 , wherein the features of the preprocessed Arabic like alphanumeric character include directional features, histogram-based features and statistical features.
3. The method of claim 2 , further comprising: validating, the classification using the histogram-based features; and refining the classification in response to determining that the classification is not valid.
4. The method of claim 2 , wherein the statistical features include at least one of a length, a width, a height, a height's variance and an aspect ratio of the preprocessed Arabic like alphanumeric character.
5. The method of claim 2 , further comprising: classifying the handwritten Arabic like alphanumeric character as the digit 0 based on the statistical features.
6. The method of claim 1 , wherein the fuzzy models are generated based on directional features obtained from training samples of handwritten Arabic like alphanumeric characters.
7. The method of claim 6 , wherein the training samples are obtained from a plurality of different users.
8. The method of claim 1 , wherein the first tolerance and the second tolerance of a fuzzy model are based on a standard deviation obtained from directional features.
9. A system for Arabic like alphanumeric character recognition, the system comprising: a alphanumeric input device; and processing circuitry configured to receive a handwritten Arabic like alphanumeric character, from the alphanumeric input device; store, in a memory, fuzzy models of a plurality of Arabic like alphanumeric characters; preprocess the handwritten Arabic like alphanumeric character; extract features from the preprocessed Arabic like alphanumeric character; compute a similarity value based on fuzzy comparisons between points of the preprocessed Arabic like alphanumeric character and the stored fuzzy models; classify the handwritten Arabic like alphanumeric character based at least in part on the similarity value; and output a classified alphanumeric character, wherein the similarity value between the preprocessed Arabic like alphanumeric character S and a fuzzy model M is calculated based on a membership value calculated by applying sim ( S , M ) = 1 N ∑ i = 1 N M v ( S ( i ) , M ( i ) ) wherein the membership value is expressed as: M v ( S ( i ) , M ( i ) ) = { 0 , d > α 1 or d < α 2 1 , β 1 ≤ d ≤ β 2 α 1 - d α 1 - β , β 1 < d ≤ α 1 d - α 2 β 2 - α 2 , α 2 ≤ d < β 2 where N is a number of points, d is a directional feature of the preprocessed Arabic like alphanumeric character S at point i, β 1 and β 2 are a first tolerance and α 1 and α 2 , are a second tolerance.
10. The system of claim 9 , wherein the features of the preprocessed Arabic like alphanumeric character include directional features, histogram-based features and statistical features.
11. The system of claim 10 , wherein the processing circuitry is further configured to: validate the classification using the histogram-based features; and refine the classification in response to determining that the classification is not valid.
12. The system of claim 10 , wherein the statistical features include at least one of a length, a width, a height, a height's variance and an aspect ratio of the preprocessed Arabic like alphanumeric character.
13. The system of claim 10 , wherein the processing circuitry is further configured to: classify the handwritten Arabic like alphanumeric character as the digit 0 based on the statistical features.
14. The system of claim 9 , wherein the fuzzy models are generated based on directional features obtained from training samples of handwritten Arabic like alphanumeric characters.
15. The system of claim 14 , wherein the training samples are obtained from a plurality of different users.
16. The system of claim 9 , wherein the first tolerance and the second tolerance of a fuzzy model are based on a standard deviation obtained from directional features.
17. A system for Arabic like alphanumeric character recognition, the system comprising: an alphanumeric input device; and processing circuitry configured to receive a handwritten Arabic like alphanumeric character, from the alphanumeric input device; store, in a memory, fuzzy models of a plurality of Arabic like alphanumeric characters; preprocess the handwritten Arabic like alphanumeric character; extract features from the preprocessed Arabic like alphanumeric character; compute a similarity value based on fuzzy comparisons between points of the preprocessed Arabic like alphanumeric character and the stored fuzzy models; classify the handwritten Arabic like alphanumeric character based at least in part on the similarity value; and output a classified alphanumeric character, wherein the features of the preprocessed Arabic like alphanumeric character include directional features, histogram-based features and statistical features.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
July 24, 2015
September 27, 2016
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.